Local-first running analysis
Compare like with like. See yourself evolve.
Cursans follows your progress across comparable workouts, easy with easy and long with long. It shows what changed, how sure it is, and where the number comes from.
The app needs no account. Your training data stays on your device unless you choose to turn on Ask Sans.
The core principle
A hard day never gets compared with an easy one.
Every run lands in one of four training families. Trends are drawn inside a family, so a change in the line means a change in you, not in the workout.
Disagree with a label? Review any run and save your own. Your correction survives re-imports.
Easy and recovery runs, compared on speed per heartbeat.
Strictly over 15 km stays Long. Harder work inside it is kept as a subtype.
Sustained work confirmed by your own pace and heart-rate evidence, not by a title guess.
Repetitions are structure. The work blocks are what get compared.
What it does
Six ways to see what training is really doing.
Each view answers one question, and each number links to the formula and the limits behind it.
Evolution across comparable runs
Aerobic efficiency, pace, heart rate, cadence, stride length and ground-contact time, tracked inside each workout family and season. The trend comes from a least-squares line through every valid run, not a comparison of two endpoints.
Tiny changes are labelled stable, so noise does not become a headline.
ProgressionSample values.
One overall aerobic-fitness state
A single VO2-equivalent estimate, pooled from the evidence across all your runs, with a confidence bar and the date of the last supporting session. Tap it to see the sessions behind it.
A quiet stretch keeps the dated previous value. It never invents a gain.
Estimate, not a lab testSample values.
Durability, quarter by quarter
How well efficiency holds from the start of a long or steady run to the end. Decoupling is read against plain bands: low up to 5%, moderate up to 8%, high above that.
Quality flags and confidence tell you when a run is too short or too messy to trust.
Long and steady runsSample run: efficiency retained by quarter, early quarter = 100. Sample values.
Race predictions from training
Predictions for 5 km, 10 km, the half and the marathon come from your training, never from an old race result. Race activities are excluded, so deleting or editing a past race does not move the prediction.
Without qualifying evidence, no time is shown.
Daniels–Gilbert equationsSample values, not a forecast for any real athlete.
Ask Sans, a coach that shows its scope
One optional entry on each training page. Before you send anything you see exactly what it will read: the page, the filter, the period. Suggestions only fill the box, and nothing is sent until you press Ask.
GPS, route names and your identity are left out of the automatic context. Conversations stay on your device and are not part of backups.
Opt in, with consentBring your history with you
Import from the files you already have. Detailed FIT, TCX and GPX runs bring heart-rate, pace, power and cadence streams. Summary files fill in the long history.
Duplicates are merged, and re-importing never erases a correction.
Your data, your deviceImport today
Direct connections
Inside the app
Small summary first. Detail when you ask.
Orange is your current block and navy is the one before it, everywhere. Screens below use sample data.
Today
Your load, one fitness number and your latest run. Nothing you did not ask for.
Compare
Pick two training blocks and see what really changed between them.
Ask Sans
Ask about the chart you are looking at. You see the scope before anything leaves your phone.
Why it is different
Built to be believed, not to flatter.
Measured, imported, estimated, inferred
The four are shown apart. Field calculations are called aerobic-efficiency or economy proxies, never laboratory running economy.
Every number shows its work
Each metric has a page with its meaning, formula, fallbacks, limits and background papers. It is written for runners, not generated on demand.
It says when it does not know
A missing workout type means zero observations, not zero fitness. Confidence counts the evidence, and no time is predicted without a supporting signal.
You are your own baseline
Progress is measured against your history in comparable conditions. Population ratings appear only as a labelled reference, never as a verdict.
Local-first by default
No account is required. Activities are stored on your device, backups are yours to keep, and one action deletes everything.
The same math on every screen
Phone, desktop and web read the same recordings through the same calculations, so a value on Today matches the value in Compare.
Where it stands
Early access is being prepared.
Cursans is in active development. Here is what is working, what is in preparation and what comes next.
Working now
- Import from FIT, TCX, GPX, CSV, Strava ZIP and Apple Health XML
- Evolution by workout family, with durability and pooled fitness
- Training-based race predictions and readinessWindows, Android and web builds
- Ask Sans with first-use consent, local backup and full delete
In preparation
- iOS build and Apple Health sync
- One-tap Strava and Intervals.icu connection
- Privacy policy and store listings
- Accessibility and localization review
Next
- Daily readiness from sleep, HRV and load
- Planned versus completed workouts
- Adaptive training plans
- Maps, routes and terrain-adjusted pace
Early access
Run. Compare. Evolve.
Leave your email and we will write when early access opens.
Cursans estimates are training-derived. They are not laboratory measurements or health assessments.